Causes of biologic therapy withdrawal in patients with rheumatic diseases: An observational cohort study from...

Biologic therapy withdrawal in inflammatory rheumatic diseases is a frequent event with direct consequences on disease control, safety, and care costs. Real-world data from southeastern Mexico remain scarce. The objective of this study was to describe the causes of biologic treatment withdrawal and to explore factors associated with drug survival in patients attending a referral rheumatology clinic in southeastern Mexico.

Observational, retrospective cohort study. The unit of analysis was the biologic treatment episode. After data cleaning, 583 episodes from 426 unique patients were included (mean 1.37 episodes/patient). Kaplan-Meier curves were constructed for treatment line and biologic class; comparisons used the log-rank test (presented descriptively given potential within-patient correlation). A multivariable Cox proportional hazards model was fitted with a cluster-robust sandwich variance estimator grouped by patient. Significance level: p < 0.05.

84.0% of patients were female; median age 51 years (IQR 41-58). Rheumatoid arthritis accounted for 76.7% of episodes and anti-TNF agents for 87.3% of biologic use. A total of 175 withdrawals (30.0%) were recorded. Among episodes with a documented reason, inefficacy was the most frequent cause (n = 134; 84.3%), followed by adverse events (n = 24; 15.1%) and non-medical causes (n = 1; 0.6%). No significant difference in drug survival by treatment line was observed (log-rank p = 0.40). Descriptive differences were observed across biologic classes (log-rank p < 0.001). In the adjusted Cox model, the anti-IL-6 class was associated with lower withdrawal risk (HR 0.14; 95% CI 0.06-0.37; p < 0.001) versus anti-TNF.

In this single-center cohort from southeastern Mexico, inefficacy was the most commonly identified medical cause of biologic withdrawal. Descriptive differences in drug survival were observed across biologic classes, with the most consistent signal in the anti-IL-6 class. These findings are hypothesis-generating and should be confirmed in larger, multicenter registries with more complete covariate adjustment.